Actually some of us still expect that we're gonna get paid interest *on top* of what we bidded :P

:blink:

Random thoughts: A lot of news recently about the sub-prime mortgage market cratering, and lenders failing. The storylines often criticize the lenders that made the loans. Is there a notion developing in the public psyche that, for defaults in high risk lending, it's the lenders who have done something wrong rather than the borrowers?

Well, there were some nasty things going on. I think especially of things like mortgage payments doubling after intro period. And it wasn't properly disclosed to the borrowers.

If the mortgage payment represents 30% of the dude's income, what do you think he's gonna do when it will be 60% of his income.

In those cases, the fault lies on the lender to a much greater extend than it does on Prosper.

That being said, I end up believing that (in many cases, not all) we're doing a service to HR borrowers by NOT funding their loans

I don't disagree that the sub-prime market was full of chicanery, and people were not aware of what they were signing up for. But our sound-bite culture might be causing some people to take away the message that, if you have a high interest rate, and you are financially you don't have to pay your loans. I like the thought that the best thing lenders might do in some circumstances is not make the loan. I will steal that, without attribution!

Isn't that one of our great cultural myths: if someone is having a hard time, it is the fault of the system / big corporations / the landlord / etc. who is taking advantage of him? The "little guy" is always honest & hardworking - and exploited by those in authority and with power. Prosper has been a great reminder for me that the little guy is not always the hero.

If I remember the statistics properly, about half of the HR loans defaulted in the first year, which was probably a good portion of the 20 some percent of total loans that defaulted. That means that out of the loans that are still current, a higher percentage of them belong to the higher credit grades. So even if the HRs and Es continue to default at the same rate that they always have been, they will have proportionally less effect on the entire prosper portfolio, so the curve should flatten out. Of course we will have to see where it will flatten out to. My guess: 33%. Any other takers?

I predict the curves will flatten at 100% :D

good point liber

good point... :(

+1

In the vein of The Price is Right, I shall have to guess 1% and hope I am the closest without going over. :rolleyes:

Previously pninen stated that default rates are so high that a great number of lenders will probably lose money. Seeing these charts go up and up each month convinces me that the pertinent question has become: will ANY lenders be making money by the end of 3 years? The other day I sorted lending stats by the oldest portfolios (of $5K+ in size), and of the 100 oldest portfolios, only 3 are earning interest rate returns of 10% or more, after an average age of only 11 months (that is, less than 1/3rd of the way into the loan terms). 15 of the portfolios are earning negative returns (as low as -20.7% (and that's not even Muleshoes)). The average ROI on these 100 oldest portfolios is approximately 3.6%, which is abysmal by any standard, and well below CD rates. What are the chances these returns will recover with the passage of time? I don't know about you, but I don't see much flattening in these graphs. I wish I could take back my 10K investment, because at this point I'm thinking I'll be lucky to get most of that money back over the 3 years, let alone make any profit. I used to think that those posting doom and gloom reports here were being melodramatic; now I think they are the Prosper realists---this is the beginning of the end for Prosper. Prosper had a great idea, but the execution is horrendous; failure is only a matter of time.

So, tell me how you really feel.

since i'm h*rny tonight i call-dibs on 69% :lol:

My guess is that it will "flatten out" at 53% 1+mo late or later at month 36. :(

where is alan's old poll about how many HR's will eventually default?

Yay, it's the first or 15th of the month! :D :lol:

-arebelspy

The 06/01/2007 update...
(IMG:http://img.villagephotos.com/p/2006-6/1187065/prosperlate-2007-06-01.gif)
See my earlier posts in this thread for an explanation of methodology. This source of this data is the Prosper performance web page.

Go pninen! And I remain optimistic on the March numbers!

whoa March and April 06

has flat line been reached?

They largest criticism of Sub-Prime lending has been from the investors who purchased the Collateralized Debt Obligations(Mortgage Bonds) that were backed by these loans. They charge that the banks lowered their standards so a Sub-Prime bond of today is not equivalent to a Sub-Prime bond of 5 years ago.(Which theoretically they should be) The Sub-Prime lenders who are going bad, had too many loans on their portfolio that started to go bad before they could sell them off fully.(The loans had to perform for 6 months after they were turned into a CDO, or the banks had to buy them back)

The secondary criticism has come from people who worry about the debt burden on families. Often times lenders would sell people a mortgage that the lender knew they borrower could not afford. They quote stats that say that 25% of the borrowers will eventually go bankrupt or have their house foreclosed upon. This would be more shocking if not for the converse... 75% of Sub-Prime mortgage holders never would have had home ownership if not for Sub-Prime lending.(Thats not exactly true, some Sub-Prime mortgages were for Rental Properties)


Foolish borrowers can often share the same landscape with unscrupulous lenders, and have overlapping responsibility for default.

06/15/2007 update...
(IMG:http://img.villagephotos.com/p/2006-6/1187065/prosperlate-2007-06-15.gif)

Well, so much for March 06 flat lining...

yep

basically we can now see after 12 months, 1/5th of all loans (20%) will be 1+ late across the website

Every line shows this either 'happened' or 'will happen'.

If it continues that 20% of all loans will be 1+mo late or worse at the end of 12 months, and if we assume 20% of the good loans 1+mo late or worse a the end of the next 12 months, and 20% of the good loans at that point will become 1+mo late or worse at the end of those 12 months, it means that 49% of all loans will be 1+mo late or worse by the time they are 36 months old.

I hope as good loans "season", they default at a lower rate so more loans survive as good loans than what my simple projection shows. Unfortunately, we aren't seeing that yet.

Except the May '06 loans -- I wonder why they have performed much better than average?

It looks like the slopes for Nov06, Dec06, Jan07, and Feb07 are a little steeper than the previous months (about 2%/mo.)

Will this trend continue for future months, or will the loans generated after the Extended-Extended data perform better?


:o

looking again, even Oct 06 is pretty steep... :huh:

I have 21 HRs (oops, damn drunken first 10 days on Prosper, pre forum)

In less than 2 months if I incl my <15 I have a 5 for 21, late
2 true lates
3 <15 and on their ways

So that means by month 4, I will have 5 out of 21 HRs 1+ late or 23.8%

So I am obviously way above average ... unfortunately my skill set was picking deadbeat HRs :) Damn forums, why were you hidden in the help link back then... curse you Prosper (shaking fist like Snoopy used to do!)

We all had high hopes, once.

I continue my optimism on March 06 - of course it will go up, but I think the slope will be less than the other months.

Has anyone done an analysis of the slopes of "AA" loans only??

It's difficult to understand how anyone can draw optimistic conclusions from viewing this graph.

I just started lending on Facebook's p2p lending site but also wanted to check out Prosper. I came across this thread -- bad debt numbers are very discouraging. Facebook forces the user to build a diversified portfolio and they say that they cut off borrowers with low credit scores. Should I expect better bad debt numbers there? In other words, is bad debt driven by mistakes of lenders who made BAD choices or is the p2p lending model just fundamentally wrong? :(

Note to NewLender777, welcome to the Prosper forums.

For those who are interested, I've update the detailed wiki page graphs linked here: http://prospers.org/wiki/Data

Scroll down to the bottom of the page to see some breakdowns by month and credit grade.

If you restrict your loans to a diversified portfolio a AA, A, B and C loans here your take similar risks as facebook. Here you have the additional benefit of more credit data, more listings to choose from, better written loan requests and over a year of history of loan payment results to research.

IMO the bad results of p2p so far are due to lenders making bad choices. This is what happens when novices try to play banker with limited info, myself included.

An additional factor is too much cash chasing too few quality loans.

Important question, but honestly, nobody knows. Lendingclub (on facebook) is very new. No performance stats will be available for many months.

What pjz says is true, to some degree. Beyond that we don't really know.

How well the loans perform depends on several things which depend on how prosper and lendingclub do their job(s).

Both prosper and lendingclub must attract borrowers. Prosper does it mostly by public relations (a continueing stream of magazine articles, etc), some advertising, and a referral program. Lendingclub does it at this point mostly by attempting to capture folks from an existing social networking site (facebook). Each company has a set of people walking in the front door (so to speak). Is the set of people who walk in the front door of company A equal to the set of people who walk in the front door at company B? Probably not. Probably one is better than the other. This is equivalent to the concept of "location" for a brick-and-mortar store. We just don't know the answer yet. Which of these companies will attract better borrowers?

Both prosper and facebook must verify the identity of borrowers and attempt to reject fraudsters. Let even a small fraction of fraudsters thru and it really hurts portfolio performance. Which one will do better at this? I have no idea.

In each case, the business hopes to attract people via the use of the "community" concept, and hopes that people attracted by virtue of a community will have a lower default rate. On prosper SO FAR this has been a failure. The prosper "group" concept has not worked out as they hoped. Prosper will (hopefully) continue to adapt and change the concept so that someday they will have a positive community affect. Lendingclub has an entirely different community concept, hoping to depend on facebook affiliiations. Will that work? No one knows.

Lendingclub tells you up front that the default rates should be lower on facebook than the historical credit reporting agency stats. They tell you as if they know, but they don't know. I find that misleading. (Yea, they used the word "typically", but I find that misleading too. What they mean by "typically" doesn't include the prosper experience. I think they may be trying to say that in some little African villages some P2P lending scheme had good default rates. The situation is so different as to make that comparison meaningless of course.) Similarly Prosper's early statements about default rates turned out to be quite misleading, as loans here are going bad much faster than the historical default rates would imply. I suspect that something similar will happen on lendingclub.

Prosper lets an auction determine the interest rate. Lendingclub has some guy behind the curtain do it for you. Which is better? What assumptions did they cook into the interest rates that the facebook executives chose for you? Did they presume that the default rates would be same as those historical cra-provided numbers? They don't say! (I'll bet they did.) Did they use a mathematical model that I agree with? Or that you agree with? They don't say! etc.

After the loans originate, and some go late, both prosper and lendingclub will have to deal with late payers. What fraction of these late loans will go on to default, and what fraction will be recovered? This depends on how well each performs their collections job. We have some data on prosper (and its bad), but we have no data on lendingclub. Perhaps they will do a much better job. On the other hand, they might do even worse!

There is a small but growing body of data on which to base a decision to invest here, and there is a zero and not yet growing body of data on which to base a decision to invest at lendingclub. Comparisons of investing performance between the two are therefore impossible at this time.

I hope lendingclub will improve their transparency over time, so we can see the stats that let us anwer these questions. Then we will have some hope of establishing a basis for comparison.

That "some P2P lending scheme" frequently involves actual real-life, pre-existing groups of borrowers. Often, a few members receive and must show a good history of paying back their portion of the loan before the other members will receive their portion. In addition, some setups hold every member of the group accountable for any member's non-payment. So, yes, comparing these successful P2P lending models to Prosper doesn't make much sense. Prosper has no idea what constitutes a meaningful and effective group lending model.

poof?

Is there information on lates/defaults based off of the day of the month the payment is due? Is first of month better than middle better than end?

As helpful as it is to see this data, Im not sure how well it applies to the investing strategy of most of the people on this forum.

What I would really like to see is a graph of the lates from each of those months filtered to include borrowers that meet certain extended credit data criteria (such as 0-2/0-10/0-2)

That would probably represent a more accurate subset of what most of us invest in.

There are some *awful* listings on Prosper that get funded.. and those will of course inflate the late loan percentages. Unfortunately since there is an endless pool of new lenders there is no real 'Financial Darwinism' to eventually eliminate lenders that make poor decisions, however I think the data presented appears scarier than reality for the majority of us.

I've always thought Prosper should allow people to pick the day of the month that they money would come out of their accounts. This would allow people to schedule their loan payments to automatically be pulled on their paydays (or the day after) to make sure they have money in the account at the time.

I agree. With the uncertain timeframe for a loan to close and be funded depending on if and how much Prosper checks, it could put the loan payment date at a possibly horrible date for the borrower if it's right before they get paid and they already have other bills set up for those times. That's why I was wondering if the lates could be tracked by dates or not.

This doesn't seem meaningful to me. When a loan is 1 month late, it has already been late during every possible day of the month. At that point prosper has already done two automatic pulls, and phoned the guy every day. If resolution were as simple as pulling again, the loan would not still be late.

However, the origination dates are in the database. If you want to do a histogram of lates by day-of-month, go for it.

I'm of the opinion that for some borrowers, a Prosper pull one day after a payroll deposit is likelier to succeed than after 10 days.

me 2

dear borrower: pay me first, before you pay yourself ;)

(IMG:http://i181.photobucket.com/albums/x202/cubbiesnextyr/chart.jpg)

Like this?

Current = Current, Paid, Payoff in Progress, Repurchased
Not Current = Everything else.

Crimeny, that was fast.
But it's kinda tough to see what percentage of all loans are delinquent.

Interesting and surprising to see so much variance in total loans, tho'. For example, the 10th is low, the 13th is high - whyfor, I wonder.

So, you want to see what percentage of loans originating on the 13th of the month are late? Ok, give me a minute.

Crappily formated, but there you go.

It goes, from worst to best:
3 84.08% 15.92%
28 86.49% 13.51%
31 86.64% 13.36%
1 86.79% 13.21%
26 87.19% 12.81%
29 87.34% 12.66%
23 87.57% 12.43%
21 87.69% 12.31%
7 88.22% 11.78%
4 88.42% 11.58%
10 88.45% 11.55%
13 88.52% 11.48%
6 88.81% 11.19%
8 88.86% 11.14%
24 88.95% 11.05%
5 88.96% 11.04%
14 89.21% 10.79%
27 89.64% 10.36%
11 89.90% 10.10%
25 89.97% 10.03%
2 90.00% 10.00%
12 90.76% 9.24%
17 90.77% 9.23%
18 90.83% 9.17%
15 91.03% 8.97%
20 91.49% 8.51%
22 91.76% 8.24%
19 91.79% 8.21%
30 91.91% 8.09%
9 91.93% 8.07%
16 93.00% 7.00%

I'm not sure if this data is telling us anything or not.
... But I'm intrigued by it in a deer-in-the-headlights kinda way.